Mutli-site Active/Active Disaster Recovery (DR) Architecture on AWS
Hi guys!
I'm researching DR on AWS and when I read the Active/Active strategy, I see that AWS notes this strategy is zero downtime. I don't understand why this is zero downtime.
For example, when customers are accessing our app in one region, when this region is down so the customers still have lost connection and then waiting for failing over to the new region. I think there is still downtime but it is very short.
Could you please explain more detail about this problem?
- Newest
- Most votes
- Most comments
It depends on the nature of the service you're providing, in particular how you're storing state and data. If the service is stateless then users can be redirected to a working region as soon as Route53 deems the failed region unhealthy. In a more realistic scenario it's likely some existing sessions might be lost or delayed since the limits of CAP will be reached, but new sessions could be served from the healthy region.
We are talking about multi-site Active/Active solution here. That means both sites are actively writing and syncing. If local copy fails, it's not about failing over for the DB but it just need a single retry from application side to read/write the data. We refer to this type of availability as continuous availability, to distinguish it from the high availability where there is brief downtime during failover.
Ideally there is no failover here as second copy in other region is active and up-to-date. e.g https://docs.aws.amazon.com/AmazonRDS/latest/AuroraUserGuide/aurora-multi-master.html
Relevant content
- asked a year ago
AWS OFFICIALUpdated 8 months ago- AWS OFFICIALUpdated 4 months ago
